3

I always struggle to understand what I can and can't do with sums. In fact, even when convergence isn't an issue, I still get confused. What can I do about this problem?

For instance, I am currently trying to show that the multiplication of (formal) polynomials is associative. By polynomial over a field $F$, let us mean a function $a : \mathbb{N} \rightarrow F$, where $F$ is a field, such that for $n$ sufficiently large it holds that $a_n = 0$. Multiplication of polynomials can be defined as follows. For all polynomials $a$ and $b$ and all $n \in \mathbb{N}$, it holds that $$(ab)_n = \sum_{i,j \in \mathbb{N}}^{i+j=n} a_i b_j.$$

So the problem is to show that for all polynomials $a$, $b$ and $c$ it holds that $(ab)c=a(bc)$. Fix any polynomials $a$, $b$ and $c$

Then $$[(ab)c]_n = \sum_{m,k \in \mathbb{N}}^{m+k=n} (ab)_m c_k = \sum_{m,k \in \mathbb{N}}^{m+k=n} (\sum_{i,j \in \mathbb{N}}^{i+j=m} a_i b_j) c_k = \sum_{m,k \in \mathbb{N}}^{m+k=n} \sum_{i,j \in \mathbb{N}}^{i+j=m} a_i b_j c_k.$$

Now I want to combine the two sums into a single sum $\displaystyle \sum_{i,j,k \in \mathbb{N}}^{i+j+k=n}$. How do I justify this?

In general, how does one justify these kinds of things?

goblin GONE
  • 67,744
  • Well, polynomials are finite sums really. It's hard to tell exactly what worries you in this question, bit, for example, if we have a summation described using twe different indexing sets, you need to be sure that the same terms of the sum appear the same number of times in both descriptions. If that is the case, then the sums are the same. – Geoff Robinson Apr 02 '13 at 05:43

1 Answers1

2

Here’s how you can do it using only the most common form of summation notation.

$$\begin{align*} [(ab)c]_n&=\sum_{m,k\in\Bbb N}^{m+k=n}(ab)_mc_k\\\\ &=\sum_{m=0}^n(ab)_mc_{n-m}\\\\ &=\sum_{m=0}^n\left(\sum_{i=0}^ma_ib_{m-i}\right)c_{n-m}\\\\ &=\sum_{m=0}^n\sum_{i=0}^ma_ib_{m-i}c_{n-m}\\\\ &=\sum_{i=0}^n\sum_{m=i}^na_ib_{m-i}c_{n-m}\\\\ &=\sum_{i=0}^na_i\sum_{m=i}^nb_{m-i}c_{n-m}\\\\ &=\sum_{i=0}^na_i\sum_{k=0}^{n-i}b_kc_{n-i-k}&&(k=m-i)\\ &=\sum_{i=0}^na_i(bc)_{n-i}\\\\ &=[a(bc)]_n \end{align*}$$

Brian M. Scott
  • 616,228